Reading Between the Tokens: Improving Preference Predictions through Mechanistic Forecasting

TL;DR

This paper introduces mechanistic forecasting, a novel approach that probes internal representations of large language models to improve human preference predictions, demonstrated through election forecasting with enhanced accuracy.

Contribution

It presents a new method leveraging internal model mechanisms for preference prediction, moving beyond surface outputs, and systematically analyzes how internal representations relate to human preferences.

Findings

Mechanistic forecasting can outperform traditional output-based methods.

Latent party-encoding components are activated by demographic and ideological cues.

Internal model representations contain systematic information about human preferences.

Abstract

Large language models are increasingly used to predict human preferences in both scientific and business endeavors, yet current approaches rely exclusively on analyzing model outputs without considering the underlying mechanisms. Using election forecasting as a test case, we introduce mechanistic forecasting, a method that demonstrates that probing internal model representations offers a fundamentally different - and sometimes more effective - approach to preference prediction. Examining over 24 million configurations across 7 models, 6 national elections, multiple persona attributes, and prompt variations, we systematically analyze how demographic and ideological information activates latent party-encoding components within the respective models. We find that leveraging this internal knowledge via mechanistic forecasting (opposed to solely relying on surface-level predictions) can improve prediction accuracy. The effects vary across demographic versus opinion-based attributes, political parties, national contexts, and models. Our findings demonstrate that the latent representational structure of LLMs contains systematic, exploitable information about human preferences, establishing a new path for using language models in social science prediction tasks.